mortality rates within various socioeconomic groups by birth ranks. Her findings are shown in Figure 5. Increasing mortality rates with declining socioeconomic levels are apparent, but equally apparent is the association between increasing mortality and family size within the various socioeconomic groups, especially in the postneonatal and early childhood periods.
Similar evidence was found in the longitudinal study of pregnancies on the island of Kauai in Hawaii, reported by Yerushalmy, et al. (31). Their data were obtained during 1953 when all women on the island of Kauai who had experienced at least one pregnancy were interviewed. Their study population included 6,039 women of a total population of around 30,000 on the island. In each interview a complete reproductive history was obtained. This portion of their study was, therefore, retrospective in nature, and the hazards of the method were well known to Yerushalmy and his colleagues. They rigorously scrutinized their findings to test both reliability and adequacy. On comparing their data with that obtained from locally recorded vital statistics, with similar data from elsewhere, and with known facts concerning variations with maternal age and parity, they concluded that it was, indeed, reliable except in regard to the reporting of early fetal deaths (under 20-weeks gestation) on the part of women over 50 years of age.
The particular findings of their study which are of interest here are shown graphically in Figure 6, which shows relative mortality rates in early and late gestation and from birth through 4 years by order of pregnancy. The important point is that at all stages, including early and late fetal periods (before and after 20 weeks' gestation) death rates are highest with the higher order pregnancies. Furthermore, when childhood mortality rates were calculated (deaths in age group 1-4 per 1,000 children who survived to age 1), the correlation with birth order was strikingly direct—almost linear. The authors state:
The pattern of childhood mortality at 1 to 4 years of age exhibits a very orderly and striking variation with order of pregnancies to an even stronger degree than any of the other indices. This indicates that the high rates for the high order of pregnancy may have an environmental origin. (31, p. 87)
Further evidence is offered by Newcombe and Tavendale (32), who compared certain factors relating to 13,556 handicapped children, child deaths, or stillbirths with control data obtained from the birth certificates of 213,353 infants born between 1953 and 1959 in British Columbia. They showed clearly, for example, that congenital malformations of various kinds were associated with maternal age rather than parity—an older woman is more likely to give birth to a child with a congenital malformation regardless of her parity. There were two general exceptions: (a) a statistically significant "in-